Object tracking using block motion estimation with adaptive Kalman estimates

Abstract

Tracking moving objects is an important research area that has significant impact in computer vision, robotics and artificial intelligence. Even security and surveillance systems have hugely benefited from the tracking research. However, the tracking of moving objects can also be challenging task because of various constraints such as moving background, occlusion, noise etc. In this paper, we propose a multiple motion based tracker whose performance can be enhanced with Kalman estimates. The tracker makes use of Kalman prediction variable whenever the tracker goes deviant or out of bound. The use of Kalman estimates produces continuous tracking estimate. The performance of the proposed tracking algorithm has been tested on standard tracking video database and the results show improvement over other state of the art existing methods.